CT for attenuation correction of stress myocardial perfusion imaging in women compared to men

CT for attenuation correction of stress myocardial perfusion imaging in women compared to men

Journal of Nuclear Cardiology Volume 15, Number 4;S15-21 Abstracts Friday, September 12, 2008 images with 4-dimensional MSPECT, and mean values were...

129KB Sizes 0 Downloads 52 Views

Journal of Nuclear Cardiology Volume 15, Number 4;S15-21

Abstracts Friday, September 12, 2008

images with 4-dimensional MSPECT, and mean values were generated for each image set. Results: Relative Myocardial Counts (% Maximum). All (n⫽42) Total LAD CFX RCA Men (n⫽18) Total LAD CFX RCA Women (n⫽24) Total LAD CFX RCA

FBP

IT

IT ⴙ AC

IT ⴙAC ⴙ SC

76.6 ⫾ 4.6 76.6 ⫾ 6.7 77.7 ⫾ 6.7 74.2 ⫾ 8.4

76.5 ⫾ 4.8 77.4 ⫾ 5.2 74.8 ⫾ 7.3 74.8 ⫾ 8.7

80.5 ⫾ 3.2*** 77.1 ⫾ 4.0 84.5 ⫾ 5.0### 83.5 ⫾ 5.1###

79.2 ⫾ 3.6# 75.8 ⫾ 4.4 84.8 ⫾ 5.3### 81.3 ⫾ 5.6##

75.8 ⫾ 4.3 76.8 ⫾ 4.2 78.0 ⫾ 6.5 70.3 ⫾ 8.6

75.7 ⫾ 3.9 78.7 ⫾ 4.7 73.9 ⫾ 6.6 70.4 ⫾ 8.8

80.3 ⫾ 3.1** 77.5 ⫾ 3.9 85.2 ⫾ 5.0** 81.6 ⫾ 6.0**

79.1 ⫾ 3.6⫹ 76.0 ⫾ 4.5 85.9 ⫾ 4.8** 79.8 ⫾ 6.5**

77.3 ⫾ 4.8 76.4 ⫾ 5.7 77.5 ⫾ 6.9 77.0 ⫾ 7.2

77.1 ⫾ 5.3 76.5 ⫾ 5.5 75.4 ⫾ 7.9 78.0 ⫾ 7.1

80.8 ⫾ 3.4* 76.9 ⫾ 4.2 83.9 ⫾ 4.9# 84.9 ⫾ 3.8**

79.4 ⫾ 3.8 75.7 ⫾ 4.4 84.0 ⫾ 5.6** 82.5 ⫾ 4.7⫹⫹

*p⬍ 0.05 vs FBP and IT; **p⬍ 0.005 vs FBP and IT; ***p⬍ 0.0001 vs FBP and IT; # p⬍ 0.02 vs FBP and IT; ##p⬍ 0.0002 vs FBP and IT; ###p⬍ 0.00002 vs FBP and IT; ⫹ p⬍ 0.05 vs IT, p ⫽ 0.05 vs FBP; ⫹⫹p⬍0.05 vs FBP, p⫽0.055 vs IT.

For total counts and for each vascular distribution, raw AC and AC⫹SC myocardial image counts were significantly increased (up to 7-fold greater on the IT⫹AC images and 8-fold greater on the IT⫹AC⫹SC images) than on conventional FBP images. Conclusions: Application of AC or AC⫹SC significantly alters both normalized and raw count distributions on SPECT myocardial perfusion images, indicating that attenuation correction with or without scatter correction must be considered when comparing individual image data to normal databases for both men and women. Moreover, the increase in myocardial image counts with AC and AC⫹SC suggests that these reconstruction techniques may benefit image quality. 15.31 MYOCARDIAL PERFUSION SPECT: REST AND STRESS IN ONE ACQUISITION BE Backus,1 FA Verburg,2 MW Konijnenberg,3 JF Verzijlbergen1 1 St Antonius Hospital, Nieuwegein, Netherlands, 2Universita¨tsklinikum, Wurzburg, Germany, 3Covidien BV, Petten, Netherlands Background: Simultaneous Dual Isotope (SDI) acquisition of 201Tl rest/ 99m Tc-sestamibi stress-myocardial perfusion single-photon emission computed tomography (MPS) is a desirable new procedure in nuclear cardiology. In this protocol 201Tl is injected at rest but imaging is performed not earlier than after exercise. Therefore one must be convinced that throughout exercise 201Tl remains distributed in an identical pattern as at rest. Before SDI can be applied clinically, 201Tl rest MPS before and after an exercise test need to be compared for equality. This study assesses the variation in washout of Thallium in normally perfused and ischemic myocardium subjected to exercise. Methods: In 102 patients 201Tl was injected in rest. Rest 201Tl MPS was performed, followed by an upright bicycle exercise-test, without injection of any tracer. Subsequently post-stress 201Tl imaging was performed (see figure). All images were corrected for attenuation and decay. Quantitative analysis of mean counts-per-pixel for each segment in a 17-segment model was done using MunichHeart. Differences between rest and post-stress 201Tl MPS were calculated. Normal segments were compared to ischemic segments. Visual analysis was performed by two independent observers scoring the 17 segments on a scale of 0-4. Results: Overall global difference between rest 201Tl and post-stress 201Tl MPS was 15.4% (⫾ 0.7% ,s.e.m.). Normal (N⫽66) and ischemic (N⫽36) patients demonstrated 16.2% (⫾ 0.7%) and 14.0% (⫾ 1.4%) (p⫽0.17) washout respectively. Quantitative analysis demonstrated no significant segmental differences between normal and ischemic myocardium. Visual assessment by two independent observers revealed a significant difference between rest 201Tl and post-stress 201Tl MPS in only one patient. The clinical diagnosis in this patient would have altered from ischemia only to infarct with ischemia. Conclusions: 201Tl post-stress MPS demonstrates significant redistribution of Thallium. This washout is global over the myocardium. The post-stress 201 Tl MPS is a reliable reflection of the rest perfusion, even in ischemic

segments. SDI acquisition of clinically applicable.

S19

201

Tl rest/99mTc-sestamibi stress-MPS is

15.32 DIAGNOSTIC ACCURACY OF HYBRID CARDIAC SPECT/CT FOR ATTENUATION CORRECTION OF STRESS MYOCARDIAL PERFUSION IMAGING IN OBESE COMPARED TO NORMAL WEIGHT PATIENTS JR Corbett, JM Cahill, JN Kritzman, JJ Meden, EP Ficaro University of Michigan, Ann Arbor, MI Background: The purpose of this study was to evaluate the diagnostic accuracy of computed tomography (CT)-based attenuation corrected (AC) single-photon emission computed tomography (SPECT) myocardial perfusion imaging compared to uncorrected SPECT (NC) in consecutive obese (body mass index [BMI] ⬎ 30) compared to normal weight patients (pts) with angiographic correlates. Methods: We studied 234 pts with recent coronary angiographic correlates. This study included 101 pts with BMI ⬍ 30 (normal group) and 133 pts with BMI ⬎ 30 (obese group). The mean weight of normal pts was 75.9 ⫾ 13.0 Kg (range 46.4-104.5 Kg) and mean weight of the obese group was 107.8 ⫾ 19.7 Kg (range 63.2-181.8 Kg); BMI averaged 25.5 ⫾ 3.2 (range 16.5-29.8) in the normal group and 37.1 ⫾ 6.97 (range 30.1-66.6) in the obese group. Imaging was performed using Siemens SYMBIA-T6 SPECT-CT imaging systems (Siemens Medical Solutions, Hoffman Estates, IL) and a stress Tc-99 sestamibi protocol. Breathhold CT acquisitions were acquired at end tidal expiration, 5-7 sec. acquisition time. SPECT images were reconstructed for attenuation correction (including scatter correction and resolution recovery) using manufacturers’ software without modification. Perfusion defects were assessed by scoring the severity and extent of perfusion defects in each of the three coronary artery distributions using the standard 17-segment model. For statistical purposes, p ⬍ 0.05 was considered significant. Results: With AC sensitivity increased from 86% to 94% in normal weight pts and from 76% to 94% in obese pts (p ⬍ 0.05). Specificity increased from 81% to 94% in normal weight pts and from 48% to 83% in obese pts (p ⬍ 0.05). Accuracy increased from 85% to 94% in normal weight pts and from 69% to 91% in obese pts (p ⬍ 0.05). Increases in sensitivity, specificity and accuracy all occurred. These improvements were larger and all were significant in the obese patient group. The normal weight group also demonstrated significant improvements in sensitivity and accuracy with a trend to increased specificity. Conclusion: CT-based attenuation correction significantly improves the diagnostic accuracy of MPI SPECT in normal weight and obese patients, but the improvements are greater in obese patients. Although significantly improved, specificity is still significantly reduced in obese compared to normal weight patients. 15.33 DIAGNOSTIC ACCURACY OF HYBRID SPECT/CT FOR ATTENUATION CORRECTION OF STRESS MYOCARDIAL PERFUSION IMAGING IN WOMEN COMPARED TO MEN JR Corbett, JM Cahill, JN Kritzman, JJ Meden, EP Ficaro University of Michigan, Ann Arbor, MI Objective: The objective of this study was to evaluate the diagnostic accuracy of computed tomography (CT)-based attenuation corrected (AC) single-photon emission computed tomography (SPECT) myocardial perfusion imaging compared to uncorrected SPECT (NC) in female compared to male patients (pts) with angiographic correlates. Methods: We studied 237 pts with recent coronary angiographic correlates including 143 consecutive male (mean age 61.2 ⫾ 10.6 and mean weight 98.9 ⫾ 21.4 Kg (range 54.5-174.1 Kg)) and 94 consecutive female patients (mean age 62.5 ⫾ 12.8 and mean weight 86.8 ⫾ 24.3 Kg (range 46.4-181.8 Kg)). Imaging was performed using Siemens SYMBIA-T6 SPECT-CT imaging systems (Siemens Medical Solutions, Hoffman Estates, IL) and a stress Tc-99 sestamibi protocol. Breathhold CT acquisitions were acquired at end tidal expiration, 5-7 sec. acquisition time. SPECT images were reconstructed for attenuation correction (including scatter correction and

S20

Abstracts Friday, September 12, 2008

resolution recovery) using manufacturers’ software without modification. Perfusion defects were assessed by scoring severity and extent of perfusion defects in each of the three coronary artery distributions using the standard 17-segment model. For statistical purposes, p ⬍ 0.05 was considered significant. Results: With attenuation correction, sensitivity increased from 78% to 93% in female pts and from 82% to 94% in male pts. Specificity increased from 57% to 82% in female pts and from 61% to 91% in male pts. Accuracy increased from 72% to 90% in female pts and from 78% to 93% in male pts. Sensitivity, specificity, and accuracy all increased significantly with AC compared to NC SPECT. The increase in specificity in male pts was slightly greater than in female pts. Conclusion: CT-based attenuation correction of myocardial perfusion images showed significant improvements in sensitivity, specificity and accuracy in both female and male patients of widely varying body habitus. Specificity improvements in female patients were not as great as in male patients. 15.34 EFFECT OF RECONSTRUCTION PARAMETERS AND ACQUISITION TIMES ON MYOCARDIAL PERFUSION DISTRIBUTION IN NORMALS EP Ficaro, JN Kritzman, JR Corbett University of Michigan Health System, Ann Arbor, MI Background: Iterative reconstruction may achieve comparable image quality with reduced number of detected counts. This study looked at differences between conventional filtered backprojection (FBP) reconstructions compared to iterative reconstructions (IR) for data acquired using standard and half-time acquisitions. Methods: A random selection of patients (n ⫽ 74, 37F and 37M) with low pre-test likelihood (ⱕ 5%) for coronary artery disease undergoing a stress-rest one day protocol were included in this study. Patients were acquired with a standard single-photon emission computed tomography (SPECT) perfusion protocol defined by American Society of Nuclear Cardiology guidelines. Half-time projection data sets were computed from the acquired data by decreasing the effective pixel intensities by 50% using a Poisson statistical model where the variance of the pixel counts is consistent with a true half-time acquisition. Images were reconstructed from both projection data sets using FBP with low pass filter and an IR algorithm incorporating 3-dimensional collimator modeling (Siemens CardioFlash, Hoffman Estates, IL). The IR algorithm is a maximum-likelihood expectation maximization using 8 ordered subsets and iterations of 4, 8, and 12 iterations to further investigate the effect of iteration number used. All of the iterative reconstructed data sets were smoothed with 9.6mm Gaussian kernel to effectively match the resolution of the FBP images. Reconstructed images were resampled into polar map format using Corridor4DM software (INVIA, Ann Arbor, MI) and gender-specific databases were compiled. Regional (9 segments) and global differences between the databases, within gender, were computed using paired Students t-test. Results: For the full-time acquisition data, statistically significant (p⬍0.05) differences were seen in 4 of 9 regions between FBP and IR databases. Between IR databases, 2 of 9 regions were different for images reconstructed with 4 vs. 8 iterations. No differences were seen between 8 and 12 iterations, suggesting that convergence is achieved with 8 iterations and not 4. In comparing full-time to half-time databases, no differences were seen with the IR algorithms while 2 or 9 regions were statistically different for FBP. Conclusions: In normal patients, IR can handle reduced imaging times with no change in the normal perfusion distribution. This was not seen with FBP. The distribution with IR is different with FBP which necessitates processing specific databases for quantifying perfusion defects. Extending this analysis to patients with known CAD for the validation of reduced-time imaging is warranted. 15.35 EFFECT OF POST-FILTERS ON LEFT VENTRICULAR EJECTION FRACTION FOR GATED CARDIAC SPECT STUDIES H Babla,1 C Bai,1 S Agili,1 RL Conwell,1 J Maddahi2 1 Digirad Corporation, Poway, CA, 2UCLA - David Geffen School of Medicine, Los Angeles, CA Background: For image reconstruction using 3-dimensional (3D) algorithms in cardiac single-photon emission computed tomography (SPECT),

Journal of Nuclear Cardiology July/August 2008

users may have different preferences of post-filters. However, different post-filters can affect the calculated end-diastolic volume (EDV) and end-systolic volume (ESV) and hence the left ventricular ejection fraction (LVEF). In this work, we quantitatively evaluate the effect of different post-filters on the calculated LVEF using a large number of patient studies. Methods: We first reconstructed gated volumes for 448 gated cardiac studies using 3D Ordered Subset Expectation Maximization (OSEM) iterative reconstruction technique with resolution recovery. We then applied two different post-filters to the gated volumes. Filter 1 was Butterworth filter with cut-off frequency⫽0.30, order⫽6 (BW0.30) and filter 2 was Butterworth filter with cut-off frequency⫽0.23, order⫽6 (BW0.23). Finally, we calculated the EDV, ESV and LVEF using Cedar-Sinai’s quantitative gated SPECT package and compared the LVEF from the two post-filters using linear correlation fitting. Results: The linear trend line through the LVEF data from filter 1 and filter 2 was y ⫽ 0.960x and correlation R2 ⫽0.947. The results showed that on average the calculated LVEF was 4% lower when using BW0.30 than when using BW0.23. Conclusions: Different post filters can change the calculated LVEF for gated cardiac SPECT studies. Users need to be aware of the significance of the difference when choosing different filters for their practice.

15.36 REPROJECTED DATA FROM RECONSTRUCTED VOLUME IMAGES CANNOT REPLACE THE ACQUIRED RAW PROJECTION DATA FOR THE PURPOSE OF QUALITY ASSURANCE WHILE DIAGNOSING CARDIAC SPECT STUDIES C Bai, RL Conwell Digirad Corporation, Poway, CA Background: The published American Society of Nuclear Cardiology guidelines for cardiac single-photon emission computed tomography (SPECT) imaging require cine review of the acquired raw projection data for diagnosis quality assurance (DQA) purposes to identify potential patient motion, tissue attenuation, bowel uptake, etc. This work is to investigate if reprojected data from reconstructed volume images can be used to replace the acquired raw projection data for DQA purposes, especially if the latter is not available. Methods: We first collected a set of conventional raw projection data by (1) simulation using mathematical phantoms modeling non-uniform attenuation and noise, (2) acquisition of a physical anthropomorphic phantom with cardiac insert on a SPECT system, and (3) obtaining patient data from clinical studies. Then, we reconstructed volume images from the raw projection data without attenuation correction (AC). Later, we reprojected the reconstructed volume images to generate reprojected data. In the reprojection process, we modeled the geometry that the original raw projection data was generated/acquired so that the reprojected data had the